We use cookies.
By using the site, you agree to our Privacy Policy.

Contract number
075-15-2021-579
Time span of the project
2021-2023

As of 01.11.2022

28
Number of staff members
25
scientific publications
1
Objects of intellectual property
General information

The project is aimed at the development and the demonstration of a new computational methodology for the design of chemical structures with specific properties augmented by possible chemical reactions. The methodology will be based on new machine learning methods developed at Dr. Tetko's laboratory. This methods will be later expanded, tuned, and tested on several types chemical reactions relying on the outstanding additional the host institution’s experience in chemical synthesis (G. A. Krestov Institute of Solution Chemistry of the Russian Academy of Sciences).

Name of the project: Computer-aided synthesis of compounds with given characteristics



Goals and objectives

The objectives of the project include:

  1. Artificial intelligence (AI) methods for the production of chemical compounds with desired characteristics.
  2. An automated system for the planning of the synthesis of the proposed compounds on the basis of preliminarily determined sets of chemical reactions.


The practical value of the study

Scientific results:

  • We have accumulated and analyzed  an array of our own experimental and literature data on the crucial physico-chemical and some biological properties of all the compounds envisioned in the research plan (di- and tetrapyrrolee compounds and their complexes with p and d elements, tetrapyrrole macrocyclic compounds with different natures and locations of substitutes in the molecule, proton-conducting ionic liquids (PILs)). On the basis of an analysis of the chemical space related to all the compounds envisioned in the research plan at the OCHEM web platform (https://ochem.eu), we have created a database of the spectral properties (the location of the maxima of absorption and emission wavelengths, the value of the molar absorption coefficient, the quantum yield of singlet oxygen, the temperatures of melting and decomposition of ionic liquids etc.) on the basis of our own an      d literature data for more than 20,000 compounds (over 40,000 depending on the experimental conditions of the  estimation of the parameter). The produced database has been used for the development of models relying on machine learning methods for high-precision forecasting of the corresponding physico-chemical properties of the compounds investigated in this project. The published models are in free access (https://ochem.eu/article/134921, http://ochem.eu/article/135195) for any user of the OCHEM  web platform.
  • Our researchers have conducted a computer modeling with the use of a wide range of machine learning methods and descriptors of all the compounds envisioned in the research plan (di- and tetrapyrrolee compounds their complexes with p and d elements, tetrapyrrole macrocyclic compounds with different natures and locations of substitutes in the molecule, PIL) for the analyzed properties (the locations of the wavelength maxima of absorption and emission, the value of the molar absorption coefficient, the quantum yield of singlet oxygen, the temperatures of melting and decomposition of ionic liquids) on the basis of the collected data. Over the course of the implementation of this part of the project we developed new models based on machine learning methods to forecast:
    1. the locations of absorption wavelength maxima of BODIPY (DNN, ChemProp, Transformer CNN) - https://ochem.eu/article/134921;
    2. the locations of the maxima of absorption wavelengths and the values of the molar absorption coefficient of the Soret band of tetrapyrrole macrocyclic compounds with various natures and locations of substitutes in the molecule (DNN, RFR, XGBOOST);
    3. the quantum yield of singlet oxygen of tetrapyrrole compounds (DNN, XGBOOST, EAGCNG, TRANSNNI, GNN GIN, ChemProp);
    4. the solvatochromic sensitivity of BODIPY;
    5. the temperatures of melting, vitrification and decomposition of ionic liquids (RFR, TRANSNNI) - http://ochem.eu/article/135195.
  • We have collected our own experimental data and literature data on one-step chemical reactions characteristic of compounds envisioned in our research plan (di- and tetrapyrrolee compounds their complexes with p and d elements, tetrapyrrole macrocyclic compounds with different natures and locations of substitutes in the molecule, PIL) for the forecasting of one-step reactions. The collected information became the foundation of the main database on the yield of the products of one-step catalytic and non-catalytic chemical reactions on the web-platform OCHEM. The database includes the values for over 12,000 reactions typical of  di- and tetrapyrrolee compounds and their complexes with p and d elements, tetrapyrrole macrocyclic compounds with different natures and locations of substitutes. During the forecasting of the parameter «reaction yield» of synthesizing macrocycles and their modification by introducing peripheral substitutes, complex-formation reactions by machine learning methods  (DNN, XGBOOST, EAGCNG, TRANSNNI) for a training set, we have produced models characterized by a standard deviation of 25 per cent.
  • We have developed original and modified existing protocols of the synthesis, purification and identification of di- and tetrapyrrolee compounds and their complexes with p and d elements, tetrapyrrole macrocyclic compounds with different natures and locations of substitutes in the molecule. The main result of the implementation of this part of the project is the synthesis, purification, and identification of more than 400 di- and tetrapyrrolee compounds and their complexes with p and d elements, tetrapyrrole macrocyclic compounds with different natures and locations of substitutes in the molecule conducted using original and modified existing  methods:
    1. 26 BODIPY molecules, of which 3 asymmetrically brominated BODIPY were produced for the first time;
    2. 37 complexes of zinc(II), copper(II), cobalt(II), nickel(II), manganese(III), indium(III), palladium(II) and gold(III) with a derivative of 5,10,15,20-meso-tetraphenylporphyrin, with β-alkyl-substituted porphyrins;
    3. 112 porphyrins, of which 8 were synthesized for the first time, derivatives of  tetraphenylporphyrin (TFP) with active groups in phenyl fragments;
    4. over 300 macrocyclic compounds with the necessary nature, number and mutual arrangement of the reaction centers for the purpose of their subsequent use as receptors of small organic molecules and for the research of their binding with RNA.

For all the synthesized compounds we researched the crucial physico-chemical properties  that were used to check the produced models based on machine learning methods, more precisely:

  • An experimental and quantum-chemical research of the crucial  physico-chemical properties of the synthesized and identified di- and tetrapyrrolee compounds and their complexes with p and d elements, tetrapyrrole macrocyclic compounds with different natures and locations of substitutes in the molecule;
  • We have developed the photophysical characteristics of dipyrrin luminophores (BODIPY/PODIPY) and the factors affecting the spectral properties of compounds in solvents of different nature;
  • The Laboratory has conducted a fluorescence imaging of biological objects using  BODIPY. In an experimental and in silico research of the processes of binding of water-soluble BODIPY with blood transport proteins (bovine serum albumin and human serum albumin) we have demonstrated that the fluorescence sensors based on water-soluble BODIPY can be used for the early diagnostics of microalbuminuria and comorbidities.
  • We have conducted an in silico screening of leading compounds possessing affinity to the selected RNA targets (the TPP riboswitch, the RNA pseudoknot of SARS-CoV-2, the SAM/SAH riboswitchь, TAU exon 10);
  • Our researchers have conducted a microbiological research to determine the minimum  suppressing concentration and the minimum bactericidal concentration of compounds synthesized within the project with respect to a number of microorganisms and an assessment of the impact of the leading compounds on the membrane potential of bacterial cells;
  • We have researched the spectral characteristics of the produced porphyrinate of zinc(II), copper(II) cobalt(II), nickel(II), manganese(III), indium (III), palladium(II) and gold(III);
  • Our researchers have designed, synthesized and identified tetrapyrrole macrocyclic compounds with the necessary nature, number and mutual arrangement of  the reaction centers for the purpose of their subsequent use as receptors of small organic molecules and for the research of their binding with RNA;
  • We have conducted an experimental research of the physico-chemical properties of the synthesized tetrapyrrole compounds and studied the reversible binding of halogenide ions and a broad range of biologically active molecules by selectively modified macrocyclic receptors;
  • The Laboratory has conducted a spectral and quantum chemical research of the sensory capability of tetrapyrrole compounds in processes of recognition of ions and small heterocyclic molecules to determine the feasibility of their use in biomedical and  technological applications. We have successfully demonstrated the possibilities of the practical use of one of the synthesized chemosensors by creating test systems  on its basis. The produced test systems are a convenient and promising method of the detection of zinc ions in water without complex instumental methods;
  • We have obtained criteria of aromaticity of porphyrins, phthalocyanines, macrcyclic compounds of the ABBB, ABBB, ААВААВ, АВВАВВ and АВАВАВ tyoe (where A is remnants of aromatic diamines and B is the pyrrole-containing fragment);
  • Our researchers have conducted an in vitro research of the antimicrobial activity of ionic liquids, in which we demonstrated that the ionic liquids considered in the project lead to  a decrease in the membrane potential of Staphylococcus aureus АТСС 29213, Escherichia coli MG1655, Pseudomonas aeruginosa АТСС 27853 cells  over 30 minutes, the same way it is observed in benzalkonium chloride. This fact allows to consider  these compounds as potential antimicrobial agents;
We have designed proton-conducting ionic liquids (PILs) as dopants of membranes for medium-temperature fuel cells by varying the structures of cations and anions determining the physico-chemical and thermal characteristics of PIL. At this stage of the implementation of the project, we used casting from a solution to produce  proton-conducting membranes based on polybenzimidazole based on a PIL.

Education and retraining of personnel:
  • We have developed and implemented education programs in the disciplines «Chemoinformatics», «Machine leading methods in chemistry».
  • Employees of the Laboratory have completed the occupational retraining program «Python development».
  • 4 employees of the Laboratory have completed internships at the international scientific center SCAMT.
  • At the international conference «Conference cluster 2021» (http://cluster.isc-ras.ru) the academic team of the laboratory organized the scientific seminar «Study of physico-chemical properties of compounds with the use of machine learning methods».
Collaborations:
  • Kazan (Volga region) Federal University (Russia): in vitro research of the antimicrobial activity of the compounds synthesized within the project.
  • Helmholtz-Center Munich (Germany), Ivanovo State University of Chemistry and Technology (Russia): joint research.
  • ITMO University (Russia): joint research, collaborative scientific events.

Hide Show full
makarov, d. m., fadeeva, y. a., shmukler, l. e., & tetko, i. v.
Beware of proper validation of models for ionic Liquids!. Journal of Molecular Liquids, 2021 (344).
ksenofontov, a. a., lukanov, m. m., bocharov, p. s., berezin, m. b., & tetko, i. v.
Deep neural network model for highly accurate prediction of BODIPYs absorption. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2022 (267).
rusanov, a. i., dmitrieva, o. a., mamardashvili, n. z., & tetko, i. v.
More Is Not Always Better: Local Models Provide Accurate Predictions of Spectral Properties of Porphyrins. International Journal of Molecular Sciences, 2022 (23, 3).
ksenofontov, a. a., lukanov, m. m., & bocharov, p. s.
Can machine learning methods accurately predict the molar absorption coefficient of different classes of dyes?. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2022 (279).
bichan, n. g., ovchenkova, e.n., ksenofontov, a.a., mozgova, v. a., gruzdev m.s., chervonova u. v., shelaev i. v., lomova t. n.
Meso-carbazole substituted porphyrin complexes: Synthesis and spectral properties according to experiment, DFT calculations and the prediction by machine learning methods, Dyes and Pigments, 2022 (204).
makarov, d. m., fadeeva, y. a., shmukler, l. e., & tetko, i. v.
Machine learning models for phase transition and decomposition temperature of ionic liquids. Journal of Molecular Liquids, 2022 (366).
makarov, d. m., fadeeva, y. a., safonova, e. a., & shmukler, l. e.
Predictive modeling of the ionic liquids antibacterial activity using machine learning. Computational Biology and Chemistry, 2022 (101).
ghosh, d., koch, u., hadian, k., sattler, m., & tetko, i. v.
Highly Accurate Filters to Flag Frequent Hitters in AlphaScreen Assays by Suggesting their Mechanism. Molecular Informatics, 2021, (41, 3).
mamardashvili, g. m., kaigorodova, e. y., lebedev, i. s., & mamardashvili, n. z.
Axial complexes of Sn (IV)-tetra (4-sulfophenyl) porphyrin with azorubine in aqueous media: fluorescent probes of local viscosity and pH indicators. Journal of Molecular Liquids, 2022 (366).
ksenofontov, a. a., bocharov, p. s., ksenofontova, k. v., & antina, e. v.
Water-Soluble BODIPY-Based fluorescent probe for BSA and HSA detection. Journal of Molecular Liquids, 2022 (345).
Other laboratories and scientists
Hosting organization
Field of studies
City
Invited researcher
Time span of the project
Laboratory «Hybrid modeling and optimization methods in complex systems»

Siberian federal University - (SibFU)

Computer and information sciences

Krasnoyarsk

Stanimirović Predrag Stefan

Serbia

2022-2024

Laboratory «Research of ultra-low-latency network technologies with ultra-high density based on the extensive use of artificial intelligence for 6G networks»

The Bonch-Bruevich Saint Petersburg State University of Telecommunications

Computer and information sciences

St. Petersburg

Abd El-Latif Ahmed Abdelrahim

Egypt

2022-2024

Laboratory for Non-linear and Microwave Photonics

Ulyanovsk State University - (USU)

Computer and information sciences

Ulyanovsk

Taylor James Roy

United Kingdom, Ireland

2021-2023